#!/usr/bin/env python3
#
# Copyright 2009-2022 David Robillard <d@drobilla.net>
# Copyright 2003-2008 Christopher Schmidt <crschmidt@crschmidt.net>
# Copyright 2005-2008 Uldis Bojars <uldis.bojars@deri.org>
# Copyright 2007-2008 Sergio Fernández <sergio.fernandez@fundacionctic.org>
# SPDX-License-Identifier: MIT
#
# Based on SpecGen:
# <http://forge.morfeo-project.org/wiki_en/index.php/SpecGen>

import datetime
import markdown
import markdown.extensions
import optparse
import os
import re
import sys
import time
import xml.sax.saxutils
import xml.dom
import xml.dom.minidom

__date__ = "2011-10-26"
__version__ = __date__.replace("-", ".")
__authors__ = """
Christopher Schmidt,
Uldis Bojars,
Sergio Fernández,
David Robillard"""
__license__ = "MIT License <http://www.opensource.org/licenses/mit>"
__contact__ = "devel@lists.lv2plug.in"

try:
    from lxml import etree

    have_lxml = True
except Exception:
    have_lxml = False

try:
    import pygments
    import pygments.lexers
    import pygments.lexers.rdf
    import pygments.formatters

    have_pygments = True
except ImportError:
    print("Error importing pygments, syntax highlighting disabled")
    have_pygments = False

try:
    import rdflib
except ImportError:
    sys.exit("Error importing rdflib")

# Global Variables
classranges = {}
classdomains = {}
linkmap = {}
spec_url = None
spec_ns_str = None
spec_ns = None
spec_pre = None
spec_bundle = None
specgendir = None
ns_list = {
    "http://purl.org/dc/terms/": "dcterms",
    "http://usefulinc.com/ns/doap#": "doap",
    "http://xmlns.com/foaf/0.1/": "foaf",
    "http://www.w3.org/2002/07/owl#": "owl",
    "http://www.w3.org/1999/02/22-rdf-syntax-ns#": "rdf",
    "http://www.w3.org/2000/01/rdf-schema#": "rdfs",
    "http://www.w3.org/2001/XMLSchema#": "xsd",
}

rdf = rdflib.Namespace("http://www.w3.org/1999/02/22-rdf-syntax-ns#")
rdfs = rdflib.Namespace("http://www.w3.org/2000/01/rdf-schema#")
owl = rdflib.Namespace("http://www.w3.org/2002/07/owl#")
lv2 = rdflib.Namespace("http://lv2plug.in/ns/lv2core#")
doap = rdflib.Namespace("http://usefulinc.com/ns/doap#")
foaf = rdflib.Namespace("http://xmlns.com/foaf/0.1/")


def findStatements(model, s, p, o):
    return model.triples([s, p, o])


def findOne(m, s, p, o):
    triples = findStatements(m, s, p, o)
    try:
        return sorted(triples)[0]
    except Exception:
        return None


def getSubject(s):
    return s[0]


def getPredicate(s):
    return s[1]


def getObject(s):
    return s[2]


def getLiteralString(s):
    return s


def isResource(n):
    return type(n) == rdflib.URIRef


def isBlank(n):
    return type(n) == rdflib.BNode


def isLiteral(n):
    return type(n) == rdflib.Literal


def niceName(uri):
    global spec_bundle
    if uri.startswith(spec_ns_str):
        return uri.replace(spec_ns_str, "")
    elif uri == str(rdfs.seeAlso):
        return "See also"

    regexp = re.compile("^(.*[/#])([^/#]+)$")
    rez = regexp.search(uri)
    if not rez:
        return uri
    pref = rez.group(1)
    if pref in ns_list:
        return ns_list.get(pref, pref) + ":" + rez.group(2)
    else:
        return uri


def termName(m, urinode):
    "Trims the namespace out of a term to give a name to the term."
    return str(urinode).replace(spec_ns_str, "")


def getLabel(m, urinode):
    statement = findOne(m, urinode, rdfs.label, None)
    if statement:
        return getLiteralString(getObject(statement))
    else:
        return ""


def linkifyCodeIdentifiers(string):
    "Add links to code documentation for identifiers like LV2_Type"

    if linkmap == {}:
        return string

    if string in linkmap.keys():
        # Exact match for complete string
        return linkmap[string]

    rgx = re.compile(
        "([^a-zA-Z0-9_:])("
        + "|".join(map(re.escape, linkmap))
        + ")([^a-zA-Z0-9_:])"
    )

    def translateCodeLink(match):
        return match.group(1) + linkmap[match.group(2)] + match.group(3)

    return rgx.sub(translateCodeLink, string)


def linkifyVocabIdentifiers(m, string, classlist, proplist, instalist):
    "Add links to vocabulary documentation for prefixed names like eg:Thing"

    rgx = re.compile("([a-zA-Z0-9_-]+):([a-zA-Z0-9_-]+)")
    namespaces = getNamespaces(m)

    def translateLink(match):
        text = match.group(0)
        prefix = match.group(1)
        name = match.group(2)
        uri = rdflib.URIRef(spec_ns + name)
        if prefix == spec_pre:
            if not (
                (classlist and uri in classlist)
                or (instalist and uri in instalist)
                or (proplist and uri in proplist)
            ):
                print("warning: Link to undefined resource <%s>\n" % text)
            return '<a href="#%s">%s</a>' % (name, name)
        elif prefix in namespaces:
            return '<a href="%s">%s</a>' % (
                namespaces[match.group(1)] + match.group(2),
                match.group(0),
            )
        else:
            return text

    return rgx.sub(translateLink, string)


def prettifyHtml(m, markup, subject, classlist, proplist, instalist):
    # Syntax highlight all C code
    if have_pygments:
        code_re = re.compile('<pre class="c-code">(.*?)</pre>', re.DOTALL)
        while True:
            code = code_re.search(markup)
            if not code:
                break
            match_str = xml.sax.saxutils.unescape(code.group(1))
            code_str = pygments.highlight(
                match_str,
                pygments.lexers.CLexer(),
                pygments.formatters.HtmlFormatter(),
            )
            markup = code_re.sub(code_str, markup, 1)

    # Syntax highlight all Turtle code
    if have_pygments:
        code_re = re.compile('<pre class="turtle-code">(.*?)</pre>', re.DOTALL)
        while True:
            code = code_re.search(markup)
            if not code:
                break
            match_str = xml.sax.saxutils.unescape(code.group(1))
            code_str = pygments.highlight(
                match_str,
                pygments.lexers.rdf.TurtleLexer(),
                pygments.formatters.HtmlFormatter(),
            )
            markup = code_re.sub(code_str, markup, 1)

    # Add links to code documentation for identifiers
    markup = linkifyCodeIdentifiers(markup)

    # Add internal links for known prefixed names
    markup = linkifyVocabIdentifiers(m, markup, classlist, proplist, instalist)

    # Transform names like #foo into links into this spec if possible
    rgx = re.compile("([ \t\n\r\f\v^]+)#([a-zA-Z0-9_-]+)")

    def translateLocalLink(match):
        text = match.group(0)
        space = match.group(1)
        name = match.group(2)
        uri = rdflib.URIRef(spec_ns + name)
        if (
            (classlist and uri in classlist)
            or (instalist and uri in instalist)
            or (proplist and uri in proplist)
        ):
            return '%s<a href="#%s">%s</a>' % (space, name, name)
        else:
            print("warning: Link to undefined resource <%s>\n" % name)
            return text

    markup = rgx.sub(translateLocalLink, markup)

    if not have_lxml:
        print("warning: No Python lxml module found, output may be invalid")
    else:
        try:
            # Parse and validate documentation as XHTML Basic 1.1
            doc = (
                """<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML Basic 1.1//EN"
                      "DTD/xhtml-basic11.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
  <head xml:lang="en" profile="profile">
    <title>Validation Skeleton Document</title>
  </head>
  <body>
"""
                + markup
                + """
  </body>
</html>"""
            )

            oldcwd = os.getcwd()
            os.chdir(specgendir)
            parser = etree.XMLParser(dtd_validation=True, no_network=True)
            etree.fromstring(doc.encode("utf-8"), parser)
        except Exception as e:
            print("Invalid documentation for %s\n%s" % (subject, e))
            line_num = 1
            for line in doc.split("\n"):
                print("%3d: %s" % (line_num, line))
                line_num += 1
        finally:
            os.chdir(oldcwd)

    return markup


def formatDoc(m, urinode, literal, classlist, proplist, instalist):
    string = getLiteralString(literal)

    if literal.datatype == lv2.Markdown:
        ext = [
            "markdown.extensions.codehilite",
            "markdown.extensions.tables",
            "markdown.extensions.def_list",
        ]

        doc = markdown.markdown(string, extensions=ext)

        # Hack to make tables valid XHTML Basic 1.1
        for tag in ["thead", "tbody"]:
            doc = doc.replace("<%s>\n" % tag, "")
            doc = doc.replace("</%s>\n" % tag, "")

        return prettifyHtml(m, doc, urinode, classlist, proplist, instalist)
    else:
        doc = xml.sax.saxutils.escape(string)
        doc = linkifyCodeIdentifiers(doc)
        doc = linkifyVocabIdentifiers(m, doc, classlist, proplist, instalist)
        return "<p>%s</p>" % doc


def getComment(m, subject, classlist, proplist, instalist):
    c = findOne(m, subject, rdfs.comment, None)
    if c:
        comment = getObject(c)
        return formatDoc(m, subject, comment, classlist, proplist, instalist)

    return ""


def getDetailedDocumentation(m, subject, classlist, proplist, instalist):
    markup = ""

    d = findOne(m, subject, lv2.documentation, None)
    if d:
        doc = getObject(d)
        if doc.datatype == lv2.Markdown:
            markup += formatDoc(
                m, subject, doc, classlist, proplist, instalist
            )
        else:
            html = getLiteralString(doc)
            markup += prettifyHtml(
                m, html, subject, classlist, proplist, instalist
            )

    return markup


def getFullDocumentation(m, subject, classlist, proplist, instalist):
    # Use rdfs:comment for first summary line
    markup = getComment(m, subject, classlist, proplist, instalist)

    # Use lv2:documentation for further details
    markup += getDetailedDocumentation(
        m, subject, classlist, proplist, instalist
    )

    return markup


def getProperty(val, first=True):
    "Return a string representing a property value in a property table"
    doc = ""
    if not first:
        doc += "<tr><th></th>"  # Empty cell in header column
    doc += "<td>%s</td></tr>\n" % val
    return doc


def endProperties(first):
    if first:
        return "</tr>"
    else:
        return ""


def rdfsPropertyInfo(term, m):
    """Generate HTML for properties: Domain, range"""
    global classranges
    global classdomains
    doc = ""

    label = getLabel(m, term)
    if label != "":
        doc += "<tr><th>Label</th><td>%s</td></tr>" % label

    # Find subPropertyOf information
    rlist = ""
    first = True
    for st in findStatements(m, term, rdfs.subPropertyOf, None):
        k = getTermLink(getObject(st), term, rdfs.subPropertyOf)
        rlist += getProperty(k, first)
        first = False
    if rlist != "":
        doc += "<tr><th>Sub-property of</th>" + rlist

    # Domain stuff
    domains = findStatements(m, term, rdfs.domain, None)
    domainsdoc = ""
    first = True
    for d in sorted(domains):
        union = findOne(m, getObject(d), owl.unionOf, None)
        if union:
            uris = parseCollection(m, getObject(union))
            for uri in uris:
                domainsdoc += getProperty(
                    getTermLink(uri, term, rdfs.domain), first
                )
                add(classdomains, uri, term)
        else:
            if not isBlank(getObject(d)):
                domainsdoc += getProperty(
                    getTermLink(getObject(d), term, rdfs.domain), first
                )
        first = False
    if len(domainsdoc) > 0:
        doc += "<tr><th>Domain</th>%s" % domainsdoc

    # Range stuff
    ranges = findStatements(m, term, rdfs.range, None)
    rangesdoc = ""
    first = True
    for r in sorted(ranges):
        union = findOne(m, getObject(r), owl.unionOf, None)
        if union:
            uris = parseCollection(m, getObject(union))
            for uri in uris:
                rangesdoc += getProperty(
                    getTermLink(uri, term, rdfs.range), first
                )
                add(classranges, uri, term)
                first = False
        else:
            if not isBlank(getObject(r)):
                rangesdoc += getProperty(
                    getTermLink(getObject(r), term, rdfs.range), first
                )
        first = False
    if len(rangesdoc) > 0:
        doc += "<tr><th>Range</th>%s" % rangesdoc

    return doc


def parseCollection(model, node):
    uris = []

    while node:
        first = findOne(model, node, rdf.first, None)
        rest = findOne(model, node, rdf.rest, None)
        if not first or not rest:
            break

        uris.append(getObject(first))
        node = getObject(rest)

    return uris


def getTermLink(uri, subject=None, predicate=None):
    uri = str(uri)
    extra = ""
    if subject is not None and predicate is not None:
        extra = 'about="%s" rel="%s" resource="%s"' % (
            str(subject),
            niceName(str(predicate)),
            uri,
        )
    if uri.startswith(spec_ns_str):
        return '<a href="#%s" %s>%s</a>' % (
            uri.replace(spec_ns_str, ""),
            extra,
            niceName(uri),
        )
    else:
        return '<a href="%s" %s>%s</a>' % (uri, extra, niceName(uri))


def owlRestrictionInfo(term, m):
    """Generate OWL restriction information for Classes"""
    restrictions = []
    for s in findStatements(m, term, rdfs.subClassOf, None):
        if findOne(m, getObject(s), rdf.type, owl.Restriction):
            restrictions.append(getObject(s))

    if not restrictions:
        return ""

    doc = "<dl>"

    for r in sorted(restrictions):
        props = findStatements(m, r, None, None)
        onProp = None
        comment = None
        for p in props:
            if getPredicate(p) == owl.onProperty:
                onProp = getObject(p)
            elif getPredicate(p) == rdfs.comment:
                comment = getObject(p)
        if onProp is not None:
            doc += "<dt>Restriction on %s</dt>\n" % getTermLink(onProp)

            prop_str = ""
            for p in findStatements(m, r, None, None):
                if (
                    getPredicate(p) == owl.onProperty
                    or getPredicate(p) == rdfs.comment
                    or (
                        getPredicate(p) == rdf.type
                        and getObject(p) == owl.Restriction
                    )
                    or getPredicate(p) == lv2.documentation
                ):
                    continue

                prop_str += getTermLink(getPredicate(p))

                if isResource(getObject(p)):
                    prop_str += " " + getTermLink(getObject(p))
                elif isLiteral(getObject(p)):
                    prop_str += " " + getLiteralString(getObject(p))

            if comment is not None:
                prop_str += "\n<div>%s</div>\n" % getLiteralString(comment)

            doc += "<dd>%s</dd>" % prop_str if prop_str else ""

    doc += "</dl>"
    return doc


def rdfsClassInfo(term, m):
    """Generate rdfs-type information for Classes: ranges, and domains."""
    global classranges
    global classdomains
    doc = ""

    label = getLabel(m, term)
    if label != "":
        doc += "<tr><th>Label</th><td>%s</td></tr>" % label

    # Find superclasses
    superclasses = set()
    for st in findStatements(m, term, rdfs.subClassOf, None):
        if not isBlank(getObject(st)):
            uri = getObject(st)
            superclasses |= set([uri])

    if len(superclasses) > 0:
        doc += "\n<tr><th>Subclass of</th>"
        first = True
        for superclass in sorted(superclasses):
            doc += getProperty(getTermLink(superclass), first)
            first = False

    # Find subclasses
    subclasses = set()
    for st in findStatements(m, None, rdfs.subClassOf, term):
        if not isBlank(getObject(st)):
            uri = getSubject(st)
            subclasses |= set([uri])

    if len(subclasses) > 0:
        doc += "\n<tr><th>Superclass of</th>"
        first = True
        for superclass in sorted(subclasses):
            doc += getProperty(getTermLink(superclass), first)
            first = False

    # Find out about properties which have rdfs:domain of t
    d = classdomains.get(str(term), "")
    if d:
        dlist = ""
        first = True
        for k in sorted(d):
            dlist += getProperty(getTermLink(k), first)
            first = False
        doc += "<tr><th>In domain of</th>%s" % dlist

    # Find out about properties which have rdfs:range of t
    r = classranges.get(str(term), "")
    if r:
        rlist = ""
        first = True
        for k in sorted(r):
            rlist += getProperty(getTermLink(k), first)
            first = False
        doc += "<tr><th>In range of</th>%s" % rlist

    return doc


def isSpecial(pred):
    """Return True if `pred` shouldn't be documented generically"""
    return pred in [
        rdf.type,
        rdfs.range,
        rdfs.domain,
        rdfs.label,
        rdfs.comment,
        rdfs.subClassOf,
        rdfs.subPropertyOf,
        lv2.documentation,
        owl.withRestrictions,
    ]


def blankNodeDesc(node, m):
    properties = findStatements(m, node, None, None)
    doc = ""
    for p in sorted(properties):
        if isSpecial(getPredicate(p)):
            continue
        doc += "<tr>"
        doc += '<td class="blankterm">%s</td>\n' % getTermLink(getPredicate(p))
        if isResource(getObject(p)):
            doc += '<td class="blankdef">%s</td>\n' % getTermLink(getObject(p))
            # getTermLink(str(getObject(p)), node, getPredicate(p))
        elif isLiteral(getObject(p)):
            doc += '<td class="blankdef">%s</td>\n' % getLiteralString(
                getObject(p)
            )
        elif isBlank(getObject(p)):
            doc += (
                '<td class="blankdef">'
                + blankNodeDesc(getObject(p), m)
                + "</td>\n"
            )
        else:
            doc += '<td class="blankdef">?</td>\n'
        doc += "</tr>"
    if doc != "":
        doc = '<table class="blankdesc">\n%s\n</table>\n' % doc
    return doc


def extraInfo(term, m):
    """Generate information about misc. properties of a term"""
    doc = ""
    properties = findStatements(m, term, None, None)
    first = True
    for p in sorted(properties):
        if isSpecial(getPredicate(p)):
            continue
        doc += "<tr><th>%s</th>\n" % getTermLink(getPredicate(p))
        if isResource(getObject(p)):
            doc += getProperty(
                getTermLink(getObject(p), term, getPredicate(p)), first
            )
        elif isLiteral(getObject(p)):
            doc += getProperty(
                linkifyCodeIdentifiers(str(getObject(p))), first
            )
        elif isBlank(getObject(p)):
            doc += getProperty(str(blankNodeDesc(getObject(p), m)), first)
        else:
            doc += getProperty("?", first)

    # doc += endProperties(first)

    return doc


def rdfsInstanceInfo(term, m):
    """Generate rdfs-type information for instances"""
    doc = ""

    label = getLabel(m, term)
    if label != "":
        doc += "<tr><th>Label</th><td>%s</td></tr>" % label

    first = True
    types = ""
    for match in sorted(findStatements(m, term, rdf.type, None)):
        types += getProperty(
            getTermLink(getObject(match), term, rdf.type), first
        )
        first = False

    if types != "":
        doc += "<tr><th>Type</th>" + types

    doc += endProperties(first)

    return doc


def owlInfo(term, m):
    """Returns an extra information that is defined about a term using OWL."""
    res = ""

    # Inverse properties ( owl:inverseOf )
    first = True
    for st in findStatements(m, term, owl.inverseOf, None):
        res += getProperty(getTermLink(getObject(st)), first)
        first = False
    if res != "":
        res += endProperties(first)
        res = "<tr><th>Inverse:</th>\n" + res

    def owlTypeInfo(term, propertyType, name):
        if findOne(m, term, rdf.type, propertyType):
            return "<tr><th>Type</th><td>%s</td></tr>\n" % name
        else:
            return ""

    res += owlTypeInfo(term, owl.DatatypeProperty, "Datatype Property")
    res += owlTypeInfo(term, owl.ObjectProperty, "Object Property")
    res += owlTypeInfo(term, owl.AnnotationProperty, "Annotation Property")
    res += owlTypeInfo(
        term, owl.InverseFunctionalProperty, "Inverse Functional Property"
    )
    res += owlTypeInfo(term, owl.SymmetricProperty, "Symmetric Property")

    return res


def isDeprecated(m, subject):
    deprecated = findOne(m, subject, owl.deprecated, None)
    return deprecated and (str(deprecated[2]).find("true") >= 0)


def docTerms(category, list, m, classlist, proplist, instalist):
    """
    A wrapper class for listing all the terms in a specific class (either
    Properties, or Classes. Category is 'Property' or 'Class', list is a
    list of term URI strings, return value is a chunk of HTML.
    """
    doc = ""
    for term in list:
        if not term.startswith(spec_ns_str):
            continue

        t = termName(m, term)
        curie = term.split(spec_ns_str[-1])[1]
        if t:
            doc += '<div class="specterm" id="%s" about="%s">' % (t, term)
        else:
            doc += '<div class="specterm" about="%s">' % term

        doc += '<h3><a href="#%s">%s</a></h3>' % (getAnchor(term), curie)
        doc += '<span class="spectermtype">%s</span>' % category

        comment = getFullDocumentation(m, term, classlist, proplist, instalist)
        is_deprecated = isDeprecated(m, term)

        doc += '<div class="spectermbody">'

        terminfo = ""
        extrainfo = ""
        if category == "Property":
            terminfo += rdfsPropertyInfo(term, m)
            terminfo += owlInfo(term, m)
        if category == "Class":
            terminfo += rdfsClassInfo(term, m)
            extrainfo += owlRestrictionInfo(term, m)
        if category == "Instance":
            terminfo += rdfsInstanceInfo(term, m)

        terminfo += extraInfo(term, m)

        if len(terminfo) > 0:  # to prevent empty list (bug #882)
            doc += '\n<table class="terminfo">%s</table>\n' % terminfo

        doc += '<div class="description">'

        if is_deprecated:
            doc += '<div class="warning">Deprecated</div>'

        if comment != "":
            doc += (
                '<div class="comment" property="rdfs:comment">%s</div>'
                % comment
            )

        doc += extrainfo

        doc += "</div>"

        doc += "</div>"
        doc += "\n</div>\n\n"

    return doc


def getShortName(uri):
    uri = str(uri)
    if "#" in uri:
        return uri.split("#")[-1]
    else:
        return uri.split("/")[-1]


def getAnchor(uri):
    uri = str(uri)
    if uri.startswith(spec_ns_str):
        return uri.replace(spec_ns_str, "").replace("/", "_")
    else:
        return getShortName(uri)


def buildIndex(m, classlist, proplist, instalist=None):
    if not (classlist or proplist or instalist):
        return ""

    head = ""
    body = ""

    def termLink(m, t):
        if str(t).startswith(spec_ns_str):
            name = termName(m, t)
            return '<a href="#%s">%s</a>' % (name, name)
        else:
            return '<a href="%s">%s</a>' % (str(t), str(t))

    if len(classlist) > 0:
        head += '<th><a href="#ref-classes" />Classes</th>'
        body += "<td><ul>"
        shown = {}
        for c in sorted(classlist):
            if c in shown:
                continue

            # Skip classes that are subclasses of classes defined in this spec
            local_subclass = False
            for p in findStatements(m, c, rdfs.subClassOf, None):
                parent = str(p[2])
                if parent.startswith(spec_ns_str):
                    local_subclass = True
            if local_subclass:
                continue

            shown[c] = True
            body += "<li>" + termLink(m, c)

            def class_tree(c):
                tree = ""
                shown[c] = True

                subclasses = []
                for s in findStatements(m, None, rdfs.subClassOf, c):
                    subclasses += [getSubject(s)]

                for s in sorted(subclasses):
                    tree += "<li>" + termLink(m, s)
                    tree += class_tree(s)
                    tree += "</li>"
                if tree != "":
                    tree = "<ul>" + tree + "</ul>"
                return tree

            body += class_tree(c)
            body += "</li>"
        body += "</ul></td>\n"

    if len(proplist) > 0:
        head += '<th><a href="#ref-properties" />Properties</th>'
        body += "<td><ul>"
        for p in sorted(proplist):
            body += "<li>%s</li>" % termLink(m, p)
        body += "</ul></td>\n"

    if instalist is not None and len(instalist) > 0:
        head += '<th><a href="#ref-instances" />Instances</th>'
        body += "<td><ul>"
        for i in sorted(instalist):
            p = getShortName(i)
            anchor = getAnchor(i)
            body += '<li><a href="#%s">%s</a></li>' % (anchor, p)
        body += "</ul></td>\n"

    if head and body:
        return """<table class="index">
<thead><tr>%s</tr></thead>
<tbody><tr>%s</tr></tbody></table>
""" % (
            head,
            body,
        )

    return ""


def add(where, key, value):
    if key not in where:
        where[key] = []
    if value not in where[key]:
        where[key].append(value)


def specInformation(m, ns):
    """
    Read through the spec (provided as a Redland model) and return classlist
    and proplist. Global variables classranges and classdomains are also filled
    as appropriate.
    """
    global classranges
    global classdomains

    # Find the class information: Ranges, domains, and list of all names.
    classtypes = [rdfs.Class, owl.Class, rdfs.Datatype]
    classlist = []
    for onetype in classtypes:
        for classStatement in findStatements(m, None, rdf.type, onetype):
            for range in findStatements(
                m, None, rdfs.range, getSubject(classStatement)
            ):
                if not isBlank(getSubject(classStatement)):
                    add(
                        classranges,
                        str(getSubject(classStatement)),
                        str(getSubject(range)),
                    )
            for domain in findStatements(
                m, None, rdfs.domain, getSubject(classStatement)
            ):
                if not isBlank(getSubject(classStatement)):
                    add(
                        classdomains,
                        str(getSubject(classStatement)),
                        str(getSubject(domain)),
                    )
            if not isBlank(getSubject(classStatement)):
                klass = getSubject(classStatement)
                if klass not in classlist and str(klass).startswith(ns):
                    classlist.append(klass)

    # Create a list of properties in the schema.
    proptypes = [
        rdf.Property,
        owl.ObjectProperty,
        owl.DatatypeProperty,
        owl.AnnotationProperty,
    ]
    proplist = []
    for onetype in proptypes:
        for propertyStatement in findStatements(m, None, rdf.type, onetype):
            prop = getSubject(propertyStatement)
            if prop not in proplist and str(prop).startswith(ns):
                proplist.append(prop)

    return classlist, proplist


def specProperty(m, subject, predicate):
    "Return a property of the spec."
    for c in findStatements(m, subject, predicate, None):
        return getLiteralString(getObject(c))
    return ""


def specProperties(m, subject, predicate):
    "Return a property of the spec."
    properties = []
    for c in findStatements(m, subject, predicate, None):
        properties += [getObject(c)]
    return properties


def specAuthors(m, subject):
    "Return an HTML description of the authors of the spec."

    subjects = [subject]
    p = findOne(m, subject, lv2.project, None)
    if p:
        subjects += [getObject(p)]

    dev = set()
    for s in subjects:
        for i in findStatements(m, s, doap.developer, None):
            for j in findStatements(m, getObject(i), foaf.name, None):
                dev.add(getLiteralString(getObject(j)))

    maint = set()
    for s in subjects:
        for i in findStatements(m, s, doap.maintainer, None):
            for j in findStatements(m, getObject(i), foaf.name, None):
                maint.add(getLiteralString(getObject(j)))

    doc = ""

    devdoc = ""
    first = True
    for d in sorted(dev):
        if not first:
            devdoc += ", "

        devdoc += f'<span class="author" property="doap:developer">{d}</span>'
        first = False
    if len(dev) == 1:
        doc += f'<tr><th class="metahead">Developer</th><td>{devdoc}</td></tr>'
    elif len(dev) > 0:
        doc += (
            f'<tr><th class="metahead">Developers</th><td>{devdoc}</td></tr>'
        )

    maintdoc = ""
    first = True
    for m in sorted(maint):
        if not first:
            maintdoc += ", "

        maintdoc += (
            f'<span class="author" property="doap:maintainer">{m}</span>'
        )
        first = False
    if len(maint):
        label = "Maintainer" if len(maint) == 1 else "Maintainers"
        doc += f'<tr><th class="metahead">{label}</th><td>{maintdoc}</td></tr>'

    return doc


def specVersion(m, subject):
    """
    Return a (minorVersion, microVersion, date) tuple
    """
    # Get the date from the latest doap release
    latest_doap_revision = ""
    latest_doap_release = None
    for i in findStatements(m, subject, doap.release, None):
        for j in findStatements(m, getObject(i), doap.revision, None):
            revision = getLiteralString(getObject(j))
            if latest_doap_revision == "" or revision > latest_doap_revision:
                latest_doap_revision = revision
                latest_doap_release = getObject(i)
    date = ""
    if latest_doap_release is not None:
        for i in findStatements(m, latest_doap_release, doap.created, None):
            date = getLiteralString(getObject(i))

    # Get the LV2 version
    minor_version = 0
    micro_version = 0
    for i in findStatements(m, None, lv2.minorVersion, None):
        minor_version = int(getLiteralString(getObject(i)))
    for i in findStatements(m, None, lv2.microVersion, None):
        micro_version = int(getLiteralString(getObject(i)))
    return (minor_version, micro_version, date)


def getInstances(model, classes, properties):
    """
    Extract all resources instanced in the ontology
    (aka "everything that is not a class or a property")
    """
    instances = []
    for c in classes:
        for i in findStatements(model, None, rdf.type, c):
            if not isResource(getSubject(i)):
                continue
            inst = getSubject(i)
            if inst not in instances and str(inst) != spec_url:
                instances.append(inst)
    for i in findStatements(model, None, rdf.type, None):
        if (
            (not isResource(getSubject(i)))
            or (getSubject(i) in classes)
            or (getSubject(i) in instances)
            or (getSubject(i) in properties)
        ):
            continue
        full_uri = str(getSubject(i))
        if full_uri.startswith(spec_ns_str):
            instances.append(getSubject(i))
    return instances


def load_tags(path, docdir):
    "Build a (symbol => URI) map from a Doxygen tag file."

    if not path or not docdir:
        return {}

    def getChildText(elt, tagname):
        "Return the content of the first child node with a certain tag name."
        for e in elt.childNodes:
            if (
                e.nodeType == xml.dom.Node.ELEMENT_NODE
                and e.tagName == tagname
            ):
                return e.firstChild.nodeValue
        return ""

    def linkTo(filename, anchor, sym):
        if anchor:
            return '<span><a href="%s/%s#%s">%s</a></span>' % (
                docdir,
                filename,
                anchor,
                sym,
            )
        else:
            return '<span><a href="%s/%s">%s</a></span>' % (
                docdir,
                filename,
                sym,
            )

    tagdoc = xml.dom.minidom.parse(path)
    root = tagdoc.documentElement
    linkmap = {}
    for cn in root.childNodes:
        if (
            cn.nodeType == xml.dom.Node.ELEMENT_NODE
            and cn.tagName == "compound"
            and cn.getAttribute("kind") != "page"
        ):
            name = getChildText(cn, "name")
            filename = getChildText(cn, "filename")
            anchor = getChildText(cn, "anchor")
            if not filename.endswith(".html"):
                filename += ".html"

            if cn.getAttribute("kind") != "group":
                linkmap[name] = linkTo(filename, anchor, name)

            prefix = ""
            if cn.getAttribute("kind") == "struct":
                prefix = name + "::"

            members = cn.getElementsByTagName("member")
            for m in members:
                mname = prefix + getChildText(m, "name")
                mafile = getChildText(m, "anchorfile")
                manchor = getChildText(m, "anchor")
                linkmap[mname] = linkTo(mafile, manchor, mname)

    return linkmap


def specgen(
    specloc,
    template_path,
    style_uri,
    docdir,
    tags,
    opts,
    instances=False,
):
    """The meat and potatoes: Everything starts here."""

    global spec_bundle
    global spec_url
    global spec_ns_str
    global spec_ns
    global spec_pre
    global ns_list
    global specgendir
    global linkmap

    spec_bundle = "file://%s/" % os.path.abspath(os.path.dirname(specloc))

    # Template
    with open(template_path, "r") as f:
        template = f.read()

    # Load code documentation link map from tags file
    linkmap = load_tags(tags, docdir)

    m = rdflib.ConjunctiveGraph()

    # RDFLib adds its own prefixes, so kludge around "time" prefix conflict
    m.namespace_manager.bind(
        "time", rdflib.URIRef("http://lv2plug.in/ns/ext/time#"), replace=True
    )

    manifest_path = os.path.join(os.path.dirname(specloc), "manifest.ttl")
    if os.path.exists(manifest_path):
        m.parse(manifest_path, format="n3")
    m.parse(specloc, format="n3")

    spec_url = getOntologyNS(m)
    spec = rdflib.URIRef(spec_url)

    # Load all seeAlso files recursively
    seeAlso = set()
    done = False
    while not done:
        done = True
        for uri in specProperties(m, spec, rdfs.seeAlso):
            if uri[:7] == "file://":
                path = uri[7:]
                if (
                    path != os.path.abspath(specloc)
                    and path.endswith("ttl")
                    and path not in seeAlso
                ):
                    seeAlso.add(path)
                    m.parse(path, format="n3")
                    done = False

    spec_ns_str = spec_url
    if spec_ns_str[-1] != "/" and spec_ns_str[-1] != "#":
        spec_ns_str += "#"

    spec_ns = rdflib.Namespace(spec_ns_str)

    namespaces = getNamespaces(m)
    keys = sorted(namespaces.keys())
    prefixes_html = "<span>"
    for i in keys:
        uri = namespaces[i]
        if uri.startswith("file:"):
            continue
        ns_list[str(uri)] = i
        if (
            str(uri) == spec_url + "#"
            or str(uri) == spec_url + "/"
            or str(uri) == spec_url
        ):
            spec_pre = i
        prefixes_html += '<a href="%s">%s</a> ' % (uri, i)
    prefixes_html += "</span>"

    if spec_pre is None:
        print("No namespace prefix for %s defined" % specloc)
        sys.exit(1)

    ns_list[spec_ns_str] = spec_pre

    classlist, proplist = specInformation(m, spec_ns_str)
    classlist = sorted(classlist)
    proplist = sorted(proplist)

    instalist = None
    if instances:
        instalist = sorted(
            getInstances(m, classlist, proplist),
            key=lambda x: getShortName(x).lower(),
        )

    azlist = buildIndex(m, classlist, proplist, instalist)

    # Generate Term HTML
    classlist = docTerms("Class", classlist, m, classlist, proplist, instalist)
    proplist = docTerms(
        "Property", proplist, m, classlist, proplist, instalist
    )
    if instances:
        instlist = docTerms(
            "Instance", instalist, m, classlist, proplist, instalist
        )

    termlist = ""
    if classlist:
        termlist += '<div class="section">'
        termlist += '<h2><a id="ref-classes" />Classes</h2>' + classlist
        termlist += "</div>"

    if proplist:
        termlist += '<div class="section">'
        termlist += '<h2><a id="ref-properties" />Properties</h2>' + proplist
        termlist += "</div>"

    if instlist:
        termlist += '<div class="section">'
        termlist += '<h2><a id="ref-instances" />Instances</h2>' + instlist
        termlist += "</div>"

    name = specProperty(m, spec, doap.name)
    title = name

    template = template.replace("@TITLE@", title)
    template = template.replace("@NAME@", name)
    template = template.replace(
        "@SHORT_DESC@", specProperty(m, spec, doap.shortdesc)
    )
    template = template.replace("@URI@", spec)
    template = template.replace("@PREFIX@", spec_pre)

    filename = os.path.basename(specloc)
    basename = os.path.splitext(filename)[0]

    template = template.replace("@STYLE_URI@", style_uri)
    template = template.replace("@PREFIXES@", str(prefixes_html))
    template = template.replace("@BASE@", spec_ns_str)
    template = template.replace("@AUTHORS@", specAuthors(m, spec))
    template = template.replace("@INDEX@", azlist)
    template = template.replace("@REFERENCE@", termlist)
    template = template.replace("@FILENAME@", filename)
    template = template.replace("@HEADER@", basename + ".h")

    mail_row = ""
    if "list_email" in opts:
        mail_row = '<tr><th>Discuss</th><td><a href="mailto:%s">%s</a>' % (
            opts["list_email"],
            opts["list_email"],
        )
        if "list_page" in opts:
            mail_row += ' <a href="%s">(subscribe)</a>' % opts["list_page"]
        mail_row += "</td></tr>"
    template = template.replace("@MAIL@", mail_row)

    version = specVersion(m, spec)  # (minor, micro, date)
    date_string = version[2]
    if date_string == "":
        date_string = "Undated"

    version_string = "%s.%s" % (version[0], version[1])
    experimental = version[0] == 0 or version[1] % 2 == 1
    if experimental:
        version_string += ' <span class="warning">EXPERIMENTAL</span>'

    if isDeprecated(m, rdflib.URIRef(spec_url)):
        version_string += ' <span class="warning">DEPRECATED</span>'

    template = template.replace("@VERSION@", version_string)

    content_links = ""
    if docdir is not None:
        content_links = '<li><a href="%s">API</a></li>' % os.path.join(
            docdir, "group__%s.html" % basename
        )

    template = template.replace("@CONTENT_LINKS@", content_links)

    docs = getDetailedDocumentation(
        m, rdflib.URIRef(spec_url), classlist, proplist, instalist
    )
    template = template.replace("@DESCRIPTION@", docs)

    now = int(os.environ.get("SOURCE_DATE_EPOCH", time.time()))
    build_date = datetime.datetime.utcfromtimestamp(now)
    template = template.replace("@DATE@", build_date.strftime("%F"))
    template = template.replace("@TIME@", build_date.strftime("%F %H:%M UTC"))

    # Validate complete output page
    try:
        oldcwd = os.getcwd()
        os.chdir(specgendir)
        etree.fromstring(
            template.replace(
                '"http://www.w3.org/MarkUp/DTD/xhtml-rdfa-1.dtd"',
                '"DTD/xhtml-rdfa-1.dtd"',
            ).encode("utf-8"),
            etree.XMLParser(dtd_validation=True, no_network=True),
        )
    except Exception as e:
        sys.stderr.write(
            "error: Validation failed for %s: %s\n" % (specloc, e)
        )
    finally:
        os.chdir(oldcwd)

    return template


def save(path, text):
    try:
        with open(path, "w") as f:
            f.write(text)
            f.flush()
    except Exception:
        e = sys.exc_info()[1]
        print('Error writing to file "' + path + '": ' + str(e))


def getNamespaces(m):
    """Return a prefix:URI dictionary of all namespaces seen during parsing"""
    nspaces = {}
    for prefix, uri in m.namespaces():
        if not re.match("default[0-9]*", prefix) and not prefix == "xml":
            # Skip silly default namespaces added by rdflib
            nspaces[prefix] = uri
    return nspaces


def getOntologyNS(m):
    ns = None
    s = findOne(m, None, rdf.type, lv2.Specification)
    if not s:
        s = findOne(m, None, rdf.type, owl.Ontology)
    if s:
        if not isBlank(getSubject(s)):
            ns = str(getSubject(s))

    if ns is None:
        sys.exit("Impossible to get ontology's namespace")
    else:
        return ns


def usage():
    script = os.path.basename(sys.argv[0])
    return "Usage: %s ONTOLOGY_TTL OUTPUT_HTML [OPTION]..." % script


def _path_from_env(variable, default):
    value = os.environ.get(variable)
    return value if value and os.path.isabs(value) else default


def _paths_from_env(variable, default):
    paths = []
    value = os.environ.get(variable)
    if value:
        paths = [p for p in value.split(os.pathsep) if os.path.isabs(p)]

    return paths if paths else default


def _data_dirs():
    return _paths_from_env(
        "XDG_DATA_DIRS", ["/usr/local/share/", "/usr/share/"]
    )


if __name__ == "__main__":
    """Ontology specification generator tool"""

    data_dir = None
    for d in _data_dirs():
        path = os.path.join(d, "lv2specgen")
        if (
            os.path.exists(os.path.join(path, "template.html"))
            and os.path.exists(os.path.join(path, "style.css"))
            and os.path.exists(os.path.join(path, "pygments.css"))
        ):
            data_dir = path
            break

    if data_dir:
        # Use installed files
        specgendir = data_dir
        default_template_path = os.path.join(data_dir, "template.html")
        default_style_dir = data_dir
    else:
        script_path = os.path.realpath(__file__)
        script_dir = os.path.dirname(os.path.realpath(__file__))
        if os.path.exists(os.path.join(script_dir, "template.html")):
            # Run from source repository
            specgendir = script_dir
            lv2_source_root = os.path.dirname(script_dir)
            default_template_path = os.path.join(script_dir, "template.html")
            default_style_dir = os.path.join(lv2_source_root, "doc", "style")
        else:
            sys.stderr.write("error: Unable to find lv2specgen data\n")
            sys.exit(-2)

    opt = optparse.OptionParser(
        usage=usage(),
        description="Write HTML documentation for an RDF ontology.",
    )
    opt.add_option(
        "--list-email",
        type="string",
        dest="list_email",
        help="Mailing list email address",
    )
    opt.add_option(
        "--list-page",
        type="string",
        dest="list_page",
        help="Mailing list info page address",
    )
    opt.add_option(
        "--template",
        type="string",
        dest="template",
        default=default_template_path,
        help="Template file for output page",
    )
    opt.add_option(
        "--style-dir",
        type="string",
        dest="style_dir",
        default=default_style_dir,
        help="Stylesheet directory path",
    )
    opt.add_option(
        "--style-uri",
        type="string",
        dest="style_uri",
        default="style.css",
        help="Stylesheet URI",
    )
    opt.add_option(
        "--docdir",
        type="string",
        dest="docdir",
        default=None,
        help="Doxygen output directory",
    )
    opt.add_option(
        "--tags",
        type="string",
        dest="tags",
        default=None,
        help="Doxygen tags file",
    )
    opt.add_option(
        "-p",
        "--prefix",
        type="string",
        dest="prefix",
        help="Specification Turtle prefix",
    )
    opt.add_option(
        "-i",
        "--instances",
        action="store_true",
        dest="instances",
        help="Document instances",
    )
    opt.add_option(
        "--copy-style",
        action="store_true",
        dest="copy_style",
        help="Copy style from template directory to output directory",
    )

    (options, args) = opt.parse_args()
    opts = vars(options)

    if len(args) < 2:
        opt.print_help()
        sys.exit(-1)

    spec_pre = options.prefix
    ontology = "file:" + str(args[0])
    output = args[1]
    docdir = options.docdir
    tags = options.tags

    out = "."
    spec = args[0]
    path = os.path.dirname(spec)
    outdir = os.path.abspath(os.path.join(out, path))

    bundle = str(outdir)
    b = os.path.basename(outdir)

    if not os.access(os.path.abspath(spec), os.R_OK):
        sys.stderr.write("error: extension %s has no %s.ttl file\n" % (b, b))
        sys.exit(1)

    # Generate spec documentation
    specdoc = specgen(
        spec,
        opts["template"],
        opts["style_uri"],
        docdir,
        tags,
        opts,
        instances=True,
    )

    # Save to HTML output file
    save(output, specdoc)

    if opts["copy_style"]:
        import shutil

        for stylesheet in ["pygments.css", "style.css"]:
            style_dir = opts["style_dir"]
            output_dir = os.path.dirname(output)
            shutil.copyfile(
                os.path.join(style_dir, stylesheet),
                os.path.join(output_dir, stylesheet),
            )